Page 206 - Safety Risk Management for Medical Devices

Verification of Risk Controls 185

the Risk Control is implemented. For a discussion on verification of effectiveness of
risk controls, see Section 22.2.

22.2 VERIFICATION OF EFFECTIVENESS

Verification of effectiveness means providing objective evidence that the Risk Control
is effective at reducing risk. Many methods can be used to establish that a Risk
Control is effective in reducing risk. For example:

• Usability testing—Summative tests with real or surrogate users to show that the
Risk Control measure reduces the risk. For example, if in a formative test using
an earlier user interface design, it was found that half of the testers were making a
particular mistake, and in the summative test we show that with the final design
only 10% of the testers make the same mistake, we can conclude that the Risk
Control was effective in reducing the likelihood of the Hazard, and thus risk.
• Clinical study—These are formal controlled studies performed on humans, typi-
cally to establish physiological effects. For example, let’s say an older version of
a pacemaker had a 10% rate of skin erosion in patients, and we have imple-
mented a new design. A clinical study shows that the new design results in less
than 5% of the test participants experiencing skin erosion. This would verify
the effectiveness of the new design in reducing the risk of skin erosion.
• Preclinical study—These are formal studies performed on animals, or cadavers.
They are used to test properties such as biocompatibility, biostability, toxicity,
and efficacy. For example, let’s say a new coating material is believed to reduce
the rate of infection of an implantable device. In a preclinical study two sam-
ples of a device, one with the new coating and one without may be implanted
in each animal. If under the same conditions, the samples with the new coating
show a lower rate of infection, this would verify the effectiveness of the new
coating in reducing the risk of infection.
• Analysis and simulation—It may be possible to verify the effectiveness of a Risk
Control by analytical and simulation means. For example, if a new algorithm in
implantable defibrillators is promised to be more effective in detection of
ventricular fibrillation (VF), we could load the old and new algorithms in two
copies of the same defibrillator and play recordings of a set of ECGs of VF in
both devices. If the device with the new algorithm detects more VF episodes
than the old algorithm, this verifies the effectiveness of the new algorithm.
• Leverage verification—In some cases the simple functionality of a Risk Control is
proof of its effectiveness in controlling risks. For example, if a fuse is used to
prevent high current flow, it can be shown in verification testing that the fuse
cuts the current flow when current reaches a certain level. That verifies that
the risk of electric shock is reduced.